Mixture-of-Subspaces in Low-Rank Adaptation
This is an incremental improvement for efficient fine-tuning of large language, multimodal, and diffusion models.
The paper tackles the problem of improving Low-Rank Adaptation (LoRA) for large models by introducing a mixture-of-subspaces approach, resulting in MoSLoRA, which consistently outperforms LoRA on tasks like commonsense reasoning, visual instruction tuning, and text-to-image generation.
In this paper, we introduce a subspace-inspired Low-Rank Adaptation (LoRA) method, which is computationally efficient, easy to implement, and readily applicable to large language, multimodal, and diffusion models. Initially, we equivalently decompose the weights of LoRA into two subspaces, and find that simply mixing them can enhance performance. To study such a phenomenon, we revisit it through a fine-grained subspace lens, showing that such modification is equivalent to employing a fixed mixer to fuse the subspaces. To be more flexible, we jointly learn the mixer with the original LoRA weights, and term the method Mixture-of-Subspaces LoRA (MoSLoRA). MoSLoRA consistently outperforms LoRA on tasks in different modalities, including commonsense reasoning, visual instruction tuning, and subject-driven text-to-image generation, demonstrating its effectiveness and robustness. Codes are available at https://github.com/wutaiqiang/MoSLoRA.